Date:
Monday , August 28, 2000
Author:
Kyle Bennett
Google +1

1.13 GHz by Intel

You might remember a rash of PIII 1.13GHz Reviews posted when the NDA was lifted a couple of weeks back. We had a 1.13GHz part to show to you guys, but we held back a review. We usually reserve "postage" for things we have done instead of things that we did not do. This time we want to explain what did happen and why. Strangely enough, it includes Tom and Anand...

continued...

Conclusions & Delusions:

I will be the first person to tell you that I am not qualified to come to a conclusion here. My experience in the PC world is shallow enough coming to a fixed conclusion and making sweeping statements about Intel's product line is out of my league. I hope that others do speak their mind on this that have more experience than me. I hope Intel comes back to us with answers as to why we are seeing any problems on the "officially supported" i820 board. The BX results are really not pertinent here since that is not a prescribed Intel platform, but they are still interesting at the least.

Looking at it another way, I would have to GUESS that something is rotten in Denmark. (all you guys in Denmark, please excuse the expression) Tom Pabst gets 100% credit for bringing Linux Kernel Compilation to light. All three CPUs failing this test was bothersome for sure. We could put a 1GHz Flip Chip with a Slocket on the same system and it would run the compilation complete without issue.

Article Image

Are there some issues with Intel's 1.13GHz part? I don't know about all of them, but from the ones I have seen and gotten to play with, none of them seem to act the same so it would seem that something is odd at best. Are they "flawed"? I am not going to go as far as to say that, but I don't think I would want the [H]ard|OCP server which runs Redhat Linux to have two PIII 1.13s stuck in it. The fact that all three CPUs failed the Linux compile would be a big red flag in my book and it seemed to be in the Intel engineer's also. He happened to keep his comments to a minimum.

The other BIG red flag is the Sysmark2000 failures we saw constantly with Tom's CPU. Check this graphic out.

Article Image

See Intel's logo stuck on there bigger than Dallas? Sysmark2000 is a "sanctioned" benchmark that Tom's CPU could not pass while the AnandTech CPU could. Tom's CPU not navigating Sysmark2000 is certainly suspect of a problem. One thing that needs to be mentioned here is that these CPUs are NOT engineering samples but rather seem to be production Intel parts that you or I could buy from a retailer. The fact that they are acting differently under the same conditions is bizarre.

The fact that my returned CPU acted totally different from my original CPU also bothered me to some extent. Two things are possible. Our testing conditions had changed considerably or the CPU was not the same one. I am not going to speculate on this. :) I told Intel I had the serial number written down, but in my hurry to ship it, I forgot. Also, I never really expected to see it again. I should have used the internal identifier, but I did not.

Bottom line is this, I really really really want to know what Intel has to say.

I want to make sure you guys know that Tom Pabst gets most of the credit for lining up the proper Linux testing and shining the light on that. I would have most certainly skipped over it. Also he questioned us to how Sysmark2000 ran and this is NOT a usual benchmark for us. (Thanks to BAPCO for helping out with that one by Intel's request.) Thanks to Anand for lending his CPU and his support. Thanks to everyone involved for allowing the HardOCP to pull everything together at a central point for evaluation.

We will let you know what becomes of this.

Again, here are the links to Tom's First and Second articles, Anand's Review, and Gamers Depot Review.

UPDATE: Tom has posted his AMD 1.1GHz TBird Reivew today. He visits the Intel issue regarding the 1.13GHz parts and Tom does have some feedback from Intel on page 3 !!